Sabre-metrics

[sjci]

We don't really have a thread dedicated to Sabres adv stats/league stats/fun numbers. So here's an attempt at a couple things

Updated @ 8PM 1/24/2015

[FIELDSET="McEichel Race"]

Rank | Team | GP | W | L | OT | PTS | ROW | P% | SORS | GR/H/A | vEDM | vBUF | vCAR | vARZ
#30 |

| 47 | 14 | 30 | 3 | 31 | 8 | .330 | .559 | 35/16/19 | 1 | - | 1 | 2
#29 |

| 47 | 12 | 26 | 9 | 31 | 9 | .351 | .563 | 35/17/18 | - | 1 | 1 | -
#28 |

| 46 | 16 | 25 | 5 | 37 | 13 | .402 | .585 | 36/17/19 | - | 2 | 1 | -
#27 |

| 46 | 16 | 25 | 5 | 37 | 13 | .402 | .567 | 36/20/17 | 1 | 1 | 1 | -

[/FIELDSET]

SORS = strength of remaining schedule (avg opponents P% for the rest of the season)
GR/H/A = games remaining/home/away
vEDM, vBUF etc = games left vs Edmonton, Buffalo, etc.​

-----------------------------

Season Outlook​

As of right now, we sit at 31 points with 8 Regulation/OT wins in 47 games.
Last season we had 33 points with 7 ROW in 47 games:

Last Season Standings at this point
30. Buffalo 47 GP - 33pts - 7 ROW
29. Edmonton 51 GP - 36pts - 13 ROW
28. Calgary 49 GP - 39pts - 13 ROW
---------------------
Current Rest Of Season Situation
16 Home Games
19 Away Games
22 Games versus the East; 13 versus the West

Sabres Season

With the Sabres final 35 games, 16 are home and 19 are away.

The Sabres home win % is .4 and they get points out of 50% of those games. (Last year it was .346/.5)
Their away Win % is .181, and they get points out of 22.7% of those games (last year it was .19/.33)

They have 22 Games Against the East (10H/12A) and 13 against the West (6H/7A)
Win % against the East = .313 with a Point % = .406
Win % against the West = .267 with a Point % = .267

So with that, basing their win and point percentage off of their remaining games and record versus the east & west, they could finish 10-22-3, for 54 points for an 96% chance of finishing 30th, and an 4% chance of finishing 29th

Using their Home and Away Win and Point percentage, they could finish 10-22-3 for 54 points for a 96% chance of finishing 30th, and a 4% chance of finishing 29th

-----------

Draft Order Probability
Finishing 30th this season will most likely mean a team around at most 60 points. For Buffalo, that would mean no better than 11 wins in their final 35 games

“Successful Tank†Finish
In order to finish 30th in the league, and guarantee the Sabres a top 2 pick, the Sabres most likely need to finish with less than 60 points. Preferably, the Sabres should finish more around 55 points or less to increase the chance.

60 Points would be a record of around 11-17-7 / 13-19-3 (Win 1 every 3 games)
55 points would be a record of around 9-20-6 / 10-21-4 (Win 1 every 4 games)
50 points (100% 30th finish) = record around 7-23-5 / 8-24-3 (Win 1 every 5 games)

Of Note: SCS currently pegs the Sabres to finish 9-19-7 (56 points), an 91% chance for 30th.

Probability % of finishing 30th – 29th – 28th
53 points = 98 - 02 - 0
54 points = 96 - 04 - 0
55 points = 94 - 06 - 0
56 points = 91 - 09 - 0
57 points = 87 - 12 - 0
58 Points = 82 – 17 – 1
59 Points = 76 – 23 – 1
60 Points = 68 – 30 – 2
61 Points = 59 – 37 – 4

“Need some lotto luck†Finish
If the Sabres were to finish the season with between 12-14 wins, a point total between 62-65, they would most likely be finishing in 29th place.

Probability % of finishing 30th – 29th – 28th
62 Points = 50 – 43 – 7
63 Points = 40 – 48 - 11
64 Points = 31 – 51 – 17
65 Points = 23 – 52 – 24

------------------

Current Predicted Standings

Right now, the predicted final standings and draft order (barring lottery results) would be:
as of 1/24/2015 @ 9:00 PM EST
Buffalo (29.8)
Edmonton (28.7)
Arizona (27.9)
Carolina (26.3)
(The number in brackets is the average finish after running a lot of simulations)

Meaning the Sabres would have a 20% shot at winning the lottery, but a guaranteed top 2 pick.

------------------------
In easy to understand terms:

Sabres at 60 or less points(11 wins or less): ~70% or Better chance of finishing 30th (Guaranteed Top 2 Pick)
Sabres at 62-65 points(Between 12 ~ 14 wins): Toss up between finishing 29th or 30th (Picking either 1st, 2nd or 3rd)
-------------------------------

Click to read more...

 

[sjci]

Buffalo Adv Stats

All Data pulled from War On Ice at 10PM 1/25/2015)
http://war-on-ice.com/hexteams.html?team=BUF&start0=20142015&end0=20142015

I wish I could put the graphs up here as well, but I can't grab the images as a separate url.

All numbers for Even Strength
Shooting Rate (Relative)
As we all know here, Buffalo is giving up a ton more shots compared to the league. To put it into perspective a little better, the offensive zone is broken up into 3 categories (Low Slot, High Slot, Outside/Point). Buffalo's % compared to league average

For
LS: .892
HS: .643
O/P: .742

Against
LS: 1.34
HS: 1.27
O/P: 1.09

The shooting % for goal(relative) rates (compared to league) tell the same story

BUF For
LS: 1.06
HS: .92
O/P: .6

Against
LS: .974
HS: .843
O/P: 1.16

1. Buffalo is actually converting better in the higher scoring low slot zone, but the lower number of shots as compared to their opposition is negating any possible gain they may receive
2. Also, Buffalo is getting killed in the O/P zone. Buffalo's defense has struggled this year, only 10 goals among the group. That low conversion rate, coupled with an almost double conversion rate against is another key factor in Buffalo losses this year.

Finally, the goal's have been hard to come by for the Sabres this year. Quality chances and such are a factor, but a low number of shots compared to the league/competition ( no matter the quality), will greatly effect the record.

BUF Success by Region
LS: 40/409 = 9.8%
HS: 16/413 = 3.9%
O/P: 7/666 = 1.1%

League Average
LS: 1234/13397 = 9.2%
HS: 791/18779 = 4.2%
O/P: 485/27689 = 1.8%

Buffalo is converting, percentage wise, more so than their opponents in the higher scoring low slot zone, and the high slot zone, however they are losing games still due to being heavily outshot. This may skew the numbers a bit, but you give a team more cracks at goal and eventually they'll go in.

-----

 

[sjci]

Buffalo has outshot their opponents 4 times this year and has a record of 3-1-0
Buffalo has been outshot in 40 games this season and has a record of 10-28-2
Buffalo is even shot wise in 3 games this season with a record of 1-1-1

----

Buffalo's 6 SO wins is tied for 2nd in the NHL
Buffalo's .852 Save % is good for 2nd in the NHL

------

Buffalo has scored 1st in 18 of their 47 games, with a record of 11-6-1
Buffalo has been scored on 1st in 29 of 47 games, with a record of 3-24-2
**BUF win % when trailing 1st is 2nd last (29th) in the NHL)

-----
2nd in the NHL in blocked shots (851)
Last in NHL on takeaways (205)
Last in NHL on faceoff % (45.3)

----
BUF in 1 Goal games (including SO): 12-6-3
BUF in 2 Goal games: 0-5
BUF in 3 Goal games: 2-19

Nearly 2/3 of our losses have been by 3 goals

Buffalo has been shutout the most in the NHL (8 times)

 

[BuzzKillington90]

THANK YOU!!!! Been missing these

 

[BakedBuffalo]

Thanks for the post. Good stuff

 

[sjci]

http://iguana.cs.toronto.edu/

NHL: Player Chemistry

Analytics even Bruins fans can understand

This chart show which players have the best offensive chemistry. Hover over to display the goal production between two players. A thicker line indicates more goals. A line that is thicker at one end indicates that player scores more of the goals when those two connect. Colours just differentiate the chords.

The line thickness represents the number of goals by each player. Thicker lines represent more goals scored. Updates after each game.

 

[sjci]

Win Probabilities

One way of figuring out a team's potential success/failure/standing is buy looking at goals, specifically the ratio of goals for and against. This type of statistical analysis in sports, known as the Pythagorean Win-Loss formula, started with baseball and runs. It has been adapted to hockey over the past few years on a pretty statistically relevant basis without mass change to the equations.

Since the end of the tie era (05-06) the team with the worst goal differential has finished last in the NHL 8 out of 9 times (07-08 last place finished with 3rd worst differential). And this year, Buffalo is leading that category by a large margin. At this rate, they will finish with a -136 goal differential!

There are a few ways to calculate win probability as applied to hockey, with varying degrees of fit to the data, however, most are a fit within 6%-7%.

Static Linear Method
Pr(win) = .5 + (GFg - GAg)/GPW

This model works well most of the time, but when you get into the extremes of data (>.700% and <.300%) you run into issues. For example, Buffalo's current win % = .298, the model however predicts a win % of .258. So based on that, Buffalo is over performing their data by 15.5%. Not crazy, but teams with very low winning percentages tend to significantly over-perform a linear prediction.

Pythagorean formula
Pr(win) = GFg^2/(GFg^2+GAg^2)

Buffalo Win % = .298
Model Prediction = .284
Differential = 4.9%

The basic formula calls for just squaring the data, however the creator of this formula, Bill James, speculated that an exponent value of less than two would be ideal. That number has been theorized for hockey to be around 1.86

(1.86) Prediction = .236
Differential = 26.3%

Pythagenport

This formula better accounts for better total scoring, with a formula of:

Pr(Win) = GFg^E / (GFg^E + GAg^E) where E = M x log(GFg + GAg) + K

This model formula predicts current win % should be around .230
------------------

If Buffalo would be closer to the average NHL team in SO this year, say 3-4 instead of 6-1, that would drop their win % to .234, that new winning % would drop the differential to the NHL theorized Pythagorean formula to a less than 1% differential. The Pythagenport formula answer (.230) would have differential of 1.7%.

So what I believe this boils down to, and many others have come to this conclusion as well, is that the Sabres have an inflated point total due to their shootout record. The Sabres should be more around a 28 point team right now. Their high save % in shootouts has currently given them 3 more points (3 more SO wins) in the standings than they should probably have.

 

[Jamie Walker]

Win Probabilities

One way of figuring out a team's potential success/failure/standing is buy looking at goals, specifically the ratio of goals for and against. This type of statistical analysis in sports, known as the Pythagorean Win-Loss formula, started with baseball and runs. It has been adapted to hockey over the past few years on a pretty statistically relevant basis without mass change to the equations.

Since the end of the tie era (05-06) the team with the worst goal differential has finished last in the NHL 8 out of 9 times (07-08 last place finished with 3rd worst differential). And this year, Buffalo is leading that category by a large margin. At this rate, they will finish with a -136 goal differential!

There are a few ways to calculate win probability as applied to hockey, with varying degrees of fit to the data, however, most are a fit within 6%-7%.

Static Linear Method
Pr(win) = .5 + (GFg - GAg)/GPW

This model works well most of the time, but when you get into the extremes of data (>.700% and <.300%) you run into issues. For example, Buffalo's current win % = .298, the model however predicts a win % of .258. So based on that, Buffalo is over performing their data by 15.5%. Not crazy, but teams with very low winning percentages tend to significantly over-perform a linear prediction.

Pythagorean formula
Pr(win) = GFg^2/(GFg^2+GAg^2)

Buffalo Win % = .298
Model Prediction = .284
Differential = 4.9%

The basic formula calls for just squaring the data, however the creator of this formula, Bill James, speculated that an exponent value of less than two would be ideal. That number has been theorized for hockey to be around 1.86

(1.86) Prediction = .236
Differential = 26.3%

Pythagenport

This formula better accounts for better total scoring, with a formula of:

Pr(Win) = GFg^E / (GFg^E + GAg^E) where E = M x log(GFg + GAg) + K

This model formula predicts current win % should be around .230
------------------

If Buffalo would be closer to the average NHL team in SO this year, say 3-4 instead of 6-1, that would drop their win % to .234, that new winning % would drop the differential to the NHL theorized Pythagorean formula to a less than 1% differential. The Pythagenport formula answer (.230) would have differential of 1.7%.

So what I believe this boils down to, and many others have come to this conclusion as well, is that the Sabres have an inflated point total due to their shootout record. The Sabres should be more around a 25 point team right now. Their high save % in shootouts has currently given them 6 more points (3 more wins) in the standings than they should probably have.

wouldn't that only be 3 points? they get a point in a SO loss. i thought that's why they came up with ROW

 

[sjci]

wouldn't that only be 3 points? they get a point in a SO loss. i thought that's why they came up with ROW

Yupp, my bad. Rushed to get this up, it would be 3 less points, not 6. So we should be at 28

 

[Jamie Walker]

let's hope it's enough for 30th. , btw, thanks for these posts, they prove the plan is working.